1. Field of the Invention
The present invention relates to a photoelectric conversion apparatus and a photoelectric conversion system having the apparatus. More particularly, the invention relates to a photoelectric conversion apparatus applicable to X-ray detectors for non-destructive examination such as medical care or internal examination, to image input units for business machines such as digital copiers, electronic blackboards, and facsimile machines, and so on, and also relates to a system having the apparatus.
2. Related Background Art
Presently, the majority of X-ray image pickup apparatus used for medical diagnosis employs a so-called film method in which X-rays are radiated to a human body, then the X-rays transmitted by the human body irradiate a fluorescent member for converting X-rays to visible light, and a film is exposed to fluorescence therefrom.
However, not only in Japan going into aging society, but also in the world, strong desires exist to improve the diagnostic efficiency in hospitals and to have higher-accuracy medical equipment. Even under such circumstances, the X-ray image pickup apparatus of the conventional film method requires a lot of time because of intervention of a development step of film before a doctor obtains a patient's X-ray image. In some incidents where the patient moves during X-ray photography and where exposure is misadjusted, photography must be carried out again. These are factors to impede an improvement in the efficiency of diagnosis and treatment in hospitals and also force a great load on the patient, which would be great hindrance against development to new medical society in future.
The demand for “digitization of X-ray image information” is increasing in the medical field these years. Once digitization is achieved, the doctor can know the patient's X-ray image information at optimum angles in real time, and the X-ray image information obtained can be recorded and managed using a medium such as a magneto-optical disk. Utilizing facsimile or another communication method or the like, the patient's X-ray image information can be sent within a short time to any hospital in the world.
In the non-destructive examination typified by examination of the inside of an object such as a building body, it is not permissible to repeat setting of various devices for X-ray photography or to repeat photography of necessary parts, either. In the case of the film method, however, whether photography of necessary parts is completed cannot be found before the end of development in such examination, either. Since expert's judgment is made after completion of development of film, it is not possible for the expert to give an instruction of photography at another in situ angle or an instruction of treatment in situ when necessary.
Accordingly, a demand is also high for acquisition of X-ray image information of desired parts in real time in this field.
Then, recently proposed in order to meet the demand for “digitization of X-ray image information” are X-ray image pickup devices using CCD solid state image sensing devices or amorphous silicon photoelectric conversion elements instead of the film.
FIG. 1 is an equivalent circuit diagram of an example of the two-dimensional photoelectric conversion apparatus applicable to such X-ray image pickup apparatus. FIG. 1 illustrates the two-dimensional photoelectric conversion apparatus of 3×3 for simplifying the description, but practical photoelectric conversion apparatus are constructed of much more bits, though depending upon the purpose of apparatus.
In FIG. 1, each of T1-1, T1-2, T1-3, T2-1, . . . , T3-3 designates a switching element, each of S1-1, S1-2, S1-3, S2-1, . . . , S3-3 a photoelectric conversion element, SR1 a shift register, SR2 a shift register, each of G1, G2, G3 a gate drive wire, each of M1, M2, M3 a signal wire, each of C1, C2, C3 an accumulating capacitors (e.g. equivalent additional capacitors added to a wiring), each of RES1, RES2, RES3 a reset switch, CRES a voltage pulse input section for reset, OP an operational amplifier, Ca an accumulated capacitance, each of U1, U2, U3 a switching element for read, each of N1, N2, N3 a gate drive wire for switching element U1 to U3, respectively, numeral 1 a photoelectric conversion circuit section, and numeral 2 a reading circuit section.
In FIG. 1, light hv incident to photoelectric conversion elements S1-1 to S3-3 is photoelectrically converted by the photoelectric conversion elements S1-1 to S3-3 to charges of photoelectric conversion signals, which are accumulated in interelectrode capacitances of the respective photoelectric conversion elements S1-1 to S3-3. These photoelectric conversion signals pass through switching elements T1-1 to T3-3 and signal wire M1 to M3 to become parallel voltage outputs. Further, they are converted to serial signals by the reading switch circuit section to be taken out to the outside.
In the structural example of the photoelectric conversion apparatus of FIG. 1, the photoelectric conversion elements of 9-bit pixels in total are separated in three rows, each including three bits. The above-stated operation is carried out in row units in order.
FIG. 2 is a timing chart to show an example of the operation of the photoelectric conversion apparatus shown in FIG. 1.
Optical information (hv) input to the photoelectric conversion elements Sl- 1 to S1-3 in the first row is photoelectrically converted to signal charges, which are accumulated in interelectrode capacitors in the respective photoelectric conversion elements of S1-1 to S1-3. After a lapse of constant accumulation time, the shift register SR1 gives a first voltage pulse for transfer to the gate driving wire G1 during a period of time T1 to switch the switching elements T1-1 to T1-3 on. This causes the signal charges accumulated in the respective interelectrode capacitors (S1-1 to S1-3) in the photoelectric conversion elements S1-1 to S1-3 to be transferred through the respective signal wires M1 to M3 to load capacitances C1 to C3, whereby potentials V1 to V3 of the respective load capacitances C1 to C3 are raised by a charge amount of signal (transfer operation).
Subsequently, the shift register SR2 successively gives voltage pulses to gate driving wires N1 to N3 to switch reading switches U1 to U3 on in order, thereby converting the signals of the first row having been transferred to the load capacitances C1 to C3 to serial signals, and, after impedance transformation by the voltage follower type operational amplifier OP, the signal of three pixels (Vout) is output to the outside of the photoelectric conversion apparatus during a period of time T3 (reading operation).
After that, a voltage pulse CRES for reset is applied to reset switches RES1 to RES3 during a period of time T2 to reset the load capacitances C1 to C3, thereby getting ready for the reading operation of the next row (reset operation).
Then the shift register SR1 successively drives the gate driving wires G2, G3, thereby outputting data of the all pixels of the photoelectric conversion elements S2-1 to S3-3 in time series.
Since the photoelectric conversion apparatus of an area type in which photosensors are arrayed two-dimensionally is generally arranged to successively perform the operations of transfer, reading, and reset in row units as described above, the image signals from the photoelectric conversion apparatus are intermittently output as shown by Vout in FIG. 2. Namely, the time necessary for reading one row is T1+T3+T2, and in the case of the two-dimensional photoelectric conversion apparatus of 3×3 shown in FIG. 1, the time of three times thereof is necessary for reading the all bits. For example, the size of the photoelectric conversion apparatus portion of the medical X-ray image pickup apparatus is said to be approximately 40 cm×40 cm necessary for the example of the X-ray image pickup apparatus for photographing the lung part. Supposing it is formed in pixel pitch of 100 μm, the total pixel number will be as huge as 4000×4000=16 million pixels. Simply assuming that the structure shown in FIG. 1 is used to perform the reading operation, the time of 4000×(T1+T2+T3) is necessary. Actually, the time necessary for T3 becomes longer, and therefore, a normal arrangement is provided with a plurality of (N) reading circuit sections to permit parallel reading scanning of N bits, thereby reading the all pixels in the time of 4000×(T1+T2+T3/N).
However, in the photoelectric conversion apparatus for successively performing the operations of from transfer through reading to reset, even with employing such structure, the time necessary for reading pixels in one line (=4000/N pixels) needs to include the transfer time T1 and reset time T2 every time of reading the pixels in each line, and therefore, the apparatus has a problem that the scanning time of photoelectric conversion, especially, with a lot of pixels was sometimes longer than expected. Especially, when the switching elements (T1-1 to T3-3) are constructed of amorphous silicon (hereinafter referred to as “a-Si”) TFTs (Thin Film Transistors) highly advantageous in respect of cost, they are not sufficient in switching performance as compared with switch elements made of single-crystal silicon, which leaves a subject of an improvement in achieving higher-speed reading of photoelectric conversion apparatus.
The load capacitors are illustrated as capacitance elements of reading capacitors C1 to C3 in FIG. 1, but practically, without a need for provision of separate elements, they are normally comprised of the interelectrode capacitances (Cgs) formed by the gate electrodes of the switching elements T1-1 to T3-3 and the electrodes on the side of signal wires M1 to M3. For example, when the signal charge of S1-1 in the first row is transferred, the capacitance of the load capacitor (reading capacitor) C1 is the sum of Cgs of the switching elements T1-1, T2-1, and T3-1 parasitic on the signal wire M1. Similarly, for example, when the signal charge of S2-2 in the second row is transferred, the capacitance of C2 is the sum of Cgs of the switching elements T1-2, T2-2, and T3-2 parasitic on the signal wire M2. In summary, whenever a signal charge of any photoelectric conversion element is transferred, the load capacitor (C1 to C3) is given by addition of three capacitances of Cgs of the switching elements. Similarly, when the two-dimensional photoelectric conversion apparatus is constructed of 4000×4000 pixels, the load capacitance of each signal line in the matrix will have the capacitance of Cgs×4000. On the other hand, when the signal charges of the load capacitances are converted to serial signals by the switching elements U1 to U3 in the reading circuit section, each signal charge is virtually transferred to the input capacitance (Ca in FIG. 1) parasitic to the input of the analog operational amplifier (OP amp) OP. When the transfer switching elements are made of a-Si, impedance transformation is achieved with little reducing the signal potential of the load capacitance, because the load capacitance of Cgs×4000>>Ca.
Also, there is a possibility of raising a problem that upon performing the transfer operation from the load capacitor (C1 to C3) to the operational amp OP side through the switching element (U1 to U3) controlled by the shift register SR2, the thermal noise occurring due to thermal agitation of carriers in the switching elements might degrade S/N of the photoelectric conversion apparatus in some cases. The effective value Vj of this thermal noise voltage is given, generally, byVj=(4KTRB)1/2 (Vrms).Here, K is the Boltzmann constant, 1.38×10−23 (J/K), T is an absolute temperature (K), and B is the frequency bandwidth (Hz) of system. Further, R is a resistance (Ω) in the case of the thermal noise occurring in a resistor. In the case of this system, it may be considered as ON resistance (Ω) of the switching elements.
Letting CL be the matrix-side capacitance (Cgs×4000) and Ca be the input capacitance on the operational amp OP side, the frequency bandwidth B can be approximated as B=1/(4R(CL∥Ca)) in the thermal noise voltage Vj=(4KTRB)1/2(Vrms), and therefore,
                    Vj        =                ⁢                              (                          4              ⁢                              KTR                /                                  (                                      4                    ⁢                                          R                      ⁡                                              (                                                  CL                          ||                          Ca                                                )                                                                              )                                                      )                                1            /            2                                                  =                ⁢                                            (                              KT                /                                  (                                      CL                    ||                    Ca                                    )                                            )                                      1              /              2                                .                    Here, CL∥Ca is series combined capacitance of CL and Ca.
Incidentally, if it is expressed as a charge amount, Qj=CV=(KT/(CL∥Ca))1/2(Vrms). Namely, the thermal noise voltage Vj occurring in such a system is determined only by the Boltzmann constant K, temperature T, and capacitance C (=CL∥Ca), which is normally called KTC noise. Unless otherwise stated, the thermal noise voltage will be called “KTC noise” hereinafter. This KTC noise is given in the simplified form of (KT/(CL∥Ca))1/2(Vrms). Since CL>>Ca, the KTC noise is determined nearly by (kT/Ca)1/2. The noise of this type can be reduced by increasing Ca, but there is limitation on increase in the capacitance formed in an Integrated circuit (IC).
Similarly, the KTC noise also occurs upon resetting the load capacitances to the reset potential by the reset switches RES1 to RES3, which raises the problem of reduction in S/N of the photoelectric conversion apparatus. This KTC noise upon reset is given by (KT/CL)1/2(V). The KTC noise occurring upon transfer and the KTC noise occurring upon reset appears as random noise of photoelectric conversion apparatus. Especially, if high-definition and high-gradation-level information is desired to obtain as in the medical X-ray image pickup apparatus, the apparatus will necessitate the photoelectric conversion apparatus with higher S/N ratios than the business machines such as copiers or electronic blackboards, and, the KTC noise could be a big problem.
In the photoelectric conversion circuit section, letting CS be the interelectrode capacitance of single photoelectric conversion element, CL be the load capacitance in the matrix signal wire, and Q be a total amount of accumulated signal charge after photoelectrically converted by the photoelectric conversion element, the signal potential V of the load capacitance CL on the matrix signal wire, after transfer by the transfer switching element, is given by V=Q/(CS+CL). Since single interelectrode capacitance CS is much smaller than the load capacitance CL composed of the 4000 interelectrode capacitances Cgs, it is practically approximated by V=Q/CL. When the switching elements having the interelectrode capacitance Cgs are made of an a-Si semiconductor thin film, individual differences will appear in capacitance values of load capacitance CL among apparatus because of dispersion in film thickness on fabrication of thin film, which would raise a problem that apparatus with high output and apparatus with low output are manufactured. In order to overcome it, upon constructing the system, such a countermeasure is taken as to add a general-purpose OP amp to adjust the gain, but the above example necessitates N general-purpose amplifiers, which will raise the cost of apparatus when also taking the adjustment process into consideration.
Also, the N reading circuit sections (ICs), especially in the equipment requiring high S/N ratios like the medical equipment, are not preferred to be located with long extension of the signal wires also in respect of an anti-noise property, but the necessary circuits are desired to be mounted near the photoelectric conversion circuit section. However, if many (N) ICs are provided, heat generation thereof will increase the temperature of the photoelectric conversion circuit section in some cases. Especially, when the switching elements are amorphous silicon TFTs, it is said that the dark current during OFF will increase, and there is a possibility of raising another problem that the heat generation of ICs could increase fixed pattern noise of photoelectric conversion apparatus.
For example, when the photoelectric conversion apparatus portion of the medical X-ray image pickup apparatus is constructed of a solid state image sensing apparatus, the noise quantity required for the whole apparatus including the photoelectric conversion elements is said to be 1/10000 or less against the dynamic range of signal if the image quality higher than that of the film method is desired to achieve. Namely, the resolution of 14 or more bits is required as performance of the A/D converter necessary for achieving the “digitization of X-ray image information.” A/D converters of 16 bits are commercially available presently, but it is a present status that the conversion speed decreases with increase of bit number, and to date there has been and is no high-speed A/D converter of 14 or more bits that can be used practically and actually in the X-ray image pickup apparatus having the photoelectric conversion apparatus of 4000×4000 pixels as described above.